! <scp> 01/08/02 reworked cases depending on Extended80 format. ! Test vectors for log10 ! ! ! The easy case: 38 d 1 0 OK 0 ! Zero: 38 d +0 0 z -H 38 d -0 0 z -H ! Infinity: 38 d +H 0 OK +H 38 d -H 0 i Q ! Negative cases: 38 d -1 0 i Q 38 d -2 0 i Q 38 d -2i2 0 i Q 38 d -4d5 0 i Q 38 d -1u1 0 i Q 38 d -1u4 0 i Q 38 d -Hd1 0 i Q 38 d -Hm1i2 0 i Q 38 d -Hm2i2 0 i Q 38 d -Hm2i4 0 i Q 38 d -E 0 i Q 38 d -Ei1 0 i Q 38 d -Ed1 0 i Q 38 d -Ep1 0 i Q 38 d -Em1 0 i Q 38 d -0i1 0 i Q 38 d -0i2 0 i Q 38 d -0i7 0 i Q ! NaN cases: 38 d Q 0 OK Q 38 d -Q 0 OK -Q ! Powers of 10 ! For now, only round-to-nearest is tested 38 =d 10 0 x 1 ! flags???? 38 =d 1000 0 x 3 ! <scp> 38 =e $73E6D1BA8323FE558C61 0 x 4000-1 ! log10(1.0e4000) 48 =d 0x73d658e3 0xab795204 0 0 x 0x406f4000 0x00000000 ! log10(1.0e250) 38 =d $3FFBCCCCCCCCCCCCCCCD 0 x -1 ! log10(0.1) 38 =d $3FF583126E978D4FDF3B 0 x -3 ! log10(0.001) ! <scp> 38 =e $0C179C3D73864F3805C0 0 x -4000-1 ! log10(1.0e-4000) 48 =d 0x0c06e93f 0x5da2824c 0 0 x 0xc06f4000 0x00000000 ! log10(1.0e-250) ! Powers of 2 ! For now, only round-to-nearest is tested ! <scp> harness appears to incorrectly convert the '$' constants below. ! <scp> Use long double derived values instead. ! <scp> 38 =e 1p1 0 x $3FFD9A209A84FBCFF799 ! <scp> 38 =e 1p10 0 x $4000C0A8C1263AC3F57F ! <scp> 38 =e 1p100 0 x $4003F0D2F16FC974F2DF-1 48 =d 0x40000000 0x00000000 0x0 0x0 x 0x3fd34413 0x509f79ff 48 =d 0x40900000 0x00000000 0x0 0x0 x 0x40081518 0x24c7587f 48 =d 0x46300000 0x00000000 0x0 0x0 x 0x403e1a5e 0x2df92e9e !38 =e 1p1000 0 x $40079683D6E5DDE917CB Makes no sense for powerpc <klh 9/30/93> !38 =e 1p10000 0 x $400ABC24CC9F55635DBE-1 Makes no sense for powerpc <klh 9/30/93> !38 =e 1p16383 0 x $400B9A1E320291BC0859 Makes no sense for powerpc <klh 9/30/93> ! <scp> 38 =e 1m1 0 x $BFFD9A209A84FBCFF799 ! <scp> 38 =e 1m10 0 x $C000C0A8C1263AC3F57F ! <scp> 38 =e 1m100 0 x $C003F0D2F16FC974F2DF-1 48 =d 0x3fe00000 0x00000000 0x0 0x0 x 0xbfd34413 0x509f79ff 48 =d 0x3f500000 0x00000000 0x0 0x0 x 0xc0081518 0x24c7587f 48 =d 0x39b00000 0x00000000 0x0 0x0 x 0xc03e1a5e 0x2df92e9e !38 =e 1m1000 0 x $C0079683D6E5DDE917CB Makes no sense for powerpc <klh 9/30/93> !38 =e 1m10000 0 x $C00ABC24CC9F55635DBE-1 Makes no sense for powerpc <klh 9/30/93> ! <scp> 38 =e E 0 x $C00B9A1E320291BC0859 ! <scp> 38 =e 0i1 0 x $C00B9AB5EA1AACA3E911 48 =d 0x00000000 0x00000001 0x0 0x0 x 0xc07434e6 0x420f4374 48 =d 0x00100000 0x00000000 0x0 0x0 x 0xc0733a71 0x46f72a42